Yes, I see your point. Deep learning, in particular, is a trade today. If we want to be generous we can call it an "experimental science"... but my perception

Yes, I see your point. Deep learning, in particular, is a trade today. If we want to be generous we can call it an "experimental science"... but my perception is that only a minority of papers in the field actually deserve that moniker. (Speaking as a deep learning practitioner with expertise in a narrow domain.)